EP0388156B1 - Natursprachenverarbeitungssystem - Google Patents

Natursprachenverarbeitungssystem Download PDF

Info

Publication number
EP0388156B1
EP0388156B1 EP90302676A EP90302676A EP0388156B1 EP 0388156 B1 EP0388156 B1 EP 0388156B1 EP 90302676 A EP90302676 A EP 90302676A EP 90302676 A EP90302676 A EP 90302676A EP 0388156 B1 EP0388156 B1 EP 0388156B1
Authority
EP
European Patent Office
Prior art keywords
grammatical
rule
sentence
category
phrase
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
EP90302676A
Other languages
English (en)
French (fr)
Other versions
EP0388156A3 (de
EP0388156A2 (de
Inventor
Yoshihiro Tokuume
Shogo Shibata
Koichi Masegi
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Canon Inc
Original Assignee
Canon Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Canon Inc filed Critical Canon Inc
Publication of EP0388156A2 publication Critical patent/EP0388156A2/de
Publication of EP0388156A3 publication Critical patent/EP0388156A3/de
Application granted granted Critical
Publication of EP0388156B1 publication Critical patent/EP0388156B1/de
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/205—Parsing
    • G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/253—Grammatical analysis; Style critique
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/268—Morphological analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/40—Processing or translation of natural language
    • G06F40/55—Rule-based translation
    • G06F40/56—Natural language generation

Definitions

  • the present invention relates to a natural language processing system and, more particularly, to a sentence structure generating device for use in a machine translation apparatus or the like.
  • a machine translation system for translation for example, from Japanese into English and an interactive response system arranged to receive a question in English or Japanese and output an answer in English or Japanese.
  • an input Japanese (or English) sentence is analyzed and the conceptual structure (semantic structure) of the Japanese (or English) sentence is determined.
  • the conceptual structure is in general expressed as a semantic network which comprises nodes representing individual concepts and arcs representing the relationships between the concepts.
  • the Japanese (or English) concept which has been obtained from the analysis of the Japanese (or English) sentence is transferred to a corresponding Japanese (or English) conceptual structure in order to compensate for the difference between the structures of the Japanese and English languages.
  • Translation into the corresponding English (or Japanese) sentence is achieved on the basis of the conceptual structure of the English (or Japanese) language.
  • the present invention pertains to the above-described process which is generally called "generation of a sentence structure".
  • FIG. 2 is a block diagram showing the general arrangement of a conventional machine translation system, a technical field related to the present invention will be explained below in further detail.
  • an input sentence is denoted by S21.
  • a morpheme analysis device 21 divides the input sentence S21 into a plurality of morphemes.
  • the morpheme analysis device 21 is required to process an agglutinative language, such as Japanese, in which no clear division is present between words which constitute a sentence, but it is not needed with respect to an inflectional language, such as English, in which clear divisions are present between words.
  • the input sentence S21 is subjected to an morpheme analysis in the morpheme analysis device 21 and delivered as a word string S22.
  • a sentence-structure analysis device 22 analyses the grammatical structure of the word string S22.
  • a semantic analysis device 23 performs an analysis at a semantic level on the basis of the phrase structure tree S23.
  • the semantic analysis device 23 outputs a semantic structure S24 as the result of its semantic analysis.
  • a transfer device 24 receives the semantic structure S24 of the input language (e.g., Japanese) analyzed by the semantic analysis device 23 and transfers it to the semantic structure S25 of the desired language (e.g., English).
  • a sentence structure generating device 25 generates a phrase structure tree S26 from the semantic structure S25 of the desired language.
  • a morpheme generating device 26 generates a translated sentence S27.
  • Fig. 3 shows in schematic form the display screen of a display (CRT) device (not shown) which is connected to the conventional translation apparatus shown in Fig. 2.
  • the illustrated display screen displays six windows.
  • the six windows display the input sentence S21, the phrase structure tree S23 of the input sentence S21, the semantic structure S24 of the input sentence S21, the semantic structure S25 of the desired language, the phrase structure tree S26 of the desired language, and the output translated sentence S27, respectively, in accordance with the sequence explained with reference to Fig. 2.
  • the natural language sentence generating device relates to, for example, a sentence structure generating device such as the sentence structure generating device 25 of Fig. 2.
  • the dictionary used in such a conventional sentence structure generating device stores generation symbols some of which indicate the groups of generation rules associated with the words of interest.
  • the generation symbols serve as pointers indicating the groups of generation rules related to the words.
  • the "generation rule” may be regarded as a production rule for providing a word string by examining each of the nodes and arcs present in a semantic network. As shown in Fig. 5, the generation rules may be carefully classified and prepared for each part of speech such as noun, intransitive verb, transitive verb, pronominal objective case and so on.
  • Each group of generation rules includes a plurality of generation rules. The sequence of applying the generation rules is determined in advance so that it determines the order of words.
  • Fig. 5 shows an example of a dictionary provided with the generation rules used in the conventional generation system.
  • " ⁇ " means that (1) if " ⁇ " is used in a condition field, it indicates that no condition is specified; (2) if " ⁇ " is used in an arc name field, it indicates that no arc name is specified; and (3) if " ⁇ " is used in a message field, it indicates that there is no message to be output.
  • the process of sentence generation starts with the node "go" to which the arc "ST" extends.
  • a rule interpreter for interpreting the generation rules examines the generation rules associated with "go” one by one.
  • the generation symbol of "go” indicates "VI”, which means that the generation rules of an intransitive verb (VI) are applied in the ascending order from the rule (1) shown in the table "INTRANSITIVE VERB (VI)" of Fig. 5.
  • the first rule (1) associated with the intransitive verb is applied.
  • the action which is assigned to the rule (1) is "out arc".
  • the term "out arc” means an arc which extends out of the corresponding node.
  • the action is "out arc"
  • the corresponding arc name indicates the type of the node pointed by the arc.
  • the type is "AGENT”.
  • the message attached to the illustrated node is "SUBJ”. Accordingly, the rule (1) will regard as AGENT the node pointed by the out arc and will generate the AGENT as a subject. Then, the process proceeds to a sub-network starting with the out arc AGENT. At this time, a message representing "SUBJ (subject)" is sent to the node "he”.
  • the message "SUBJ (subject)" matches the condition of the rule (5) other than the rules (1) to (4). Accordingly, as shown as “OUTPUT ONESELF” in the column "ACTION” of the generation rule (5), the word “he” itself is output.
  • the generation rule (5) of Fig. 5 does not indicate generation of a new arc, and the dictionary of Fig. 5 does not include any other generation rule subsequent to the rule (5). Accordingly, the generation of the sub-network of "he” which starts with AGENT ends with the rule (5). The process returns from the processing of the sub-network to the processing of the node "go".
  • the rules (2) et seq. of the generation rules of the intransitive verb of Fig. 5 are examined in sequence. None of the generation rules (2), (3) and (4) is applied to this processing because of the structure of the node "go". Since the arc "PAST" which represents the past tense extends out of the node "go”, the rule (5) is applied. By the application of the rule (5), the past form “went” of "go” is output. Neither of the rules (6) and (7) is applied. The process proceeds to the rule (8), according to which the processing of the out arc "GOAL" is executed.
  • ⁇ ( ⁇ 1 ... ⁇ n) is a phrase structure rule for converting ⁇ into the sequence ⁇ 1 ... ⁇ n.
  • NP represents a nominal phrase
  • VP a verbal phrase
  • A an agentive case
  • O an objective case
  • R a recipient case
  • ⁇ a predicate.
  • each phrase structure rule and the corresponding case information are described in the same part. In other words, each phrase and its meaning are described in pair in the same part.
  • the conventional sentence structure generation system provided with the above-described sentence structure generation process involves a number of problems. For example, since case information representing semantic information needs to be mixed with phrase structure rules representing sentence structure rules, it is necessary to write a plurality of identical phrase structure rules in the following manner.
  • a natural language sentence generating apparatus for generating a phrase structure from a sentence input to the apparatus, the apparatus comprising:
  • An embodiment of the present invention provides a sentence structure generating device which enables a description of grammatical rules based on a clear construction and which excels in, for example, maintainability and expandability.
  • An embodiment of the present invention provides a sentence structure generating device which can assure the grammatical correctness of a generated sentence.
  • An embodiment of the present invention provides a sentence structure generating device which can describe analysis and generation, which have heretofore been separately described, in a similar manner by means of unification grammar.
  • phrase structure tree phrase structure
  • the grammatical rule stored in the grammatical rule storing means is described by using the feature description rule.
  • the conceptual structure of the input sentence is a feature structure converted from a semantic network.
  • the apparatus further comprises analysis means for analyzing the input sentence in accordance with unification grammar and output the resulting feature structure.
  • the aforesaid generating means may include recursive call means for executing a program which can be recursively called and the conceptual structure of the sentence input by the input means is the feature structure.
  • the phrase structure part includes a phrase as a category, the phrase structure part comprising a superordinate category and at least one subordinate category constituting the superordinate category.
  • the feature structure propagated to the subordinate category may comprise an apparatus, wherein the feature structure propagated to the subordinate category comprises at least one feature structure of the superordinate category.
  • the searching means selects at least the message and, as a subordinate category, a category which matches the applying condition described in the condition part.
  • the applying condition which is described in the condition part is the type of verb or the type of clause.
  • feature is a key word in the following explanation. Each feature includes a feature name and a feature value and represents a grammatical function. Simple examples of the feature names are “subject” (“SUBJ”), "object” (“OBJ”), “gender” (“GENDER”), and “number” (“NUMBER”), "case” (“CASE”) and so on.
  • the sentence structure generating device receives as its input a feature structure which represents a feature as an element and outputs a corresponding phrase structure. It will be apparent that the introduction of the feature structure makes it possible to describe the analysis of sentence structure and the generation of sentence structure, both of which have heretofore been separately performed, in the same manner by means of the unification grammar.
  • the unification grammar describes grammar by using features and represents the relations between constituent elements of a sentence as the unification of the features, as described in (1) Schieber: An Introduction to Unification-Based approaches to Grammar, CSLI Lecture Notes No. 4. 1986 and (2) Nomura: Fundamental Technique of Natural Language Processing, compiled by the Institute of Electronics, Information and Communication Engineers and published by Corona-sha (1988).
  • the unification is one kind of sum-set operation.
  • LFG uses as its grammatical rules phrase structure rules to which are added grammatical functions.
  • the sentence structure of a sentence is represented by two hierarchies: a C-structure representing a phrase structure tree and an F-structure representing the hierarchical structure of grammatical function.
  • the C-structure (constituent structure) is a structure tree itself.
  • the C-structure represents as an analysis tree obtained by performing normal sentence analysis by using the phrase structure rules contained in the LFG grammatical rules.
  • the F-structure (feature structure) is a feature structure in which the relations between the grammatical functions described in the grammatical rules are hierarchically represented in correspondence with the sentence structure represented by the C-structure.
  • the C-structure is used in the process of obtaining the F-structure. Accordingly, the analysis result in LFG is represented with the F-structure.
  • each phrase structure rule ER shown in Fig. 6A is used as the phrase structure rule of the grammatical rules
  • the C-structure of the sentence "He sees tables.” is as shown in Fig. 6B.
  • the F-structure which results from the analysis of LFG corresponding to the C-structure is, for example, as shown in Fig. 6C.
  • the F-structure (feature structure) shown in Fig. 6C is as follows.
  • the subject "SUB” has a singular form and is represented as "NUM SG". Since the subject has the third person, the person is represented as "PERS 3".
  • the gender of the subject is represented as "GEN MASC” because of its male gender. Since the subject has a nominative case, the case is represented as "CASE NOM”. Since the content (predicate) of the subject is "HE", the predicate is represented as "PRED HE". Since the tense of this sentence is the present tense, it is represented as "TENSE PRES".
  • the predicate is represented as "SEES ⁇ (SUB)(OBJ)>".
  • SEES The predicative portion (predicate) of this sentence is "SEES” and accompanies a predicate and an object. Therefore, “SEES” is represented as “SEES ⁇ (SUB) (OBJ)>". Also, the object “OBJ” has a plural form and is represented as “NUM PL”. Its content (predicate) is "TABLE” and is represented as “PRED TABLE”.
  • the grammatical rules of LFG are represented by phrase structure rules and the grammatical functions of non-terminal symbols appearing in the phrase structure rules.
  • the grammatical functions are represented in a form called a functional schema using metavariables.
  • the grammatical rules of LFG are represented as, for instance, the six examples (LR1 to LR6) shown in Fig. 7.
  • LR1 (S ⁇ NP VP) is a phrase structure rule.
  • the rule LR1 indicates that the sentence S includes a nominal phrase NP and a verbal phrase VP.
  • An upward arrow and a downward arrow in each functional schema are called metavariables.
  • the grammatical rules of LFG include the phrase structure rules and the functional schemas.
  • terminal symbols are literal terms (for example, the above-described "he") which appear in the final process of phrase generation
  • non-terminal symbols are symbols other than terminal symbols Specifically, in S ⁇ Np VP, NP and VP are the non-terminal symbols.
  • the metavariable ⁇ indicates an F-structure which corresponds to the non-terminal symbol ("S" in the example LR1) on the left side of each phrase structure rule, that is to say, the F-structure of a superordinate node in a phrase structure tree corresponding to the phrase structure rule.
  • ( ⁇ SUB) indicates that NP to which ( ⁇ SUB) is attached is SUB of the superordinate node S in the F-structure of the superordinate node S.
  • the metavariable ⁇ indicates an F-structure which corresponds to a non-terminal symbol to which the functional schema is attached, that is to say, the F-structure of the subordinate node (or node N of Fig. 6B) of a phrase structure tree (for example, Fig.
  • the phrase structure rule 6B corresponding to the phrase structure rule.
  • the non-terminal symbol in parentheses may be omitted in each phrase structure rule.
  • the nominal phrase NP or the prepositional phrase PP may be omitted.
  • Each element marked by ⁇ may or may not appear.
  • the number of elements may be one or more, and the order of appearance does not matter.
  • the prepositional phrase PP may appear by a plurality of times. If a plurality of elements marked by ⁇ appear, the respective elements are handled independently. In other words, the elements have independent F-structure.
  • LR1 in Fig. 7 corresponds to the phrase structure rule ER1 of Fig. 6A.
  • LR1 indicates that the nominal phrase NP and the verbal phrase VP are arranged in that order in the sentence S, and each non-terminal symbol indicates that the condition shown by the functional schema attached to the non-terminal schema must be satisfied.
  • the first term of LR1 on the right side reads "F-structure of NP is shifted to the portion of SUB of the F-structure of the superordinate node.”
  • the F-structure of the nominal phrase is the F-structure of the subject SUB of the sentence S.
  • the grammatical function of the subject of the sentence is the grammatical function performed by NP in the phrase structure rule of LR1.
  • the second term of LR1 on the right side reads "F-structure of VP is shifted to the F-structure of the superordinate node.”
  • the functional schema ⁇ ⁇ indicates that, in the subordinate nodes of the phrase structure rule on the right side, a node accompanying that functional schema is a head.
  • the vocabulary entries of the dictionary are represented by using the functional schemas, as in the case of the above-described grammatical rules.
  • the metavariable ⁇ of the functional schemas refers to the F-structure of a superordinate node, as in the case of the grammatical rules described above.
  • the superordinate nodes of the vocabulary entries always represent pre-terminal symbols, respectively.
  • the metavariable ⁇ refers to the F-structure of a superordinate node, as in the case of the grammatical rules described above.
  • the superordinate node of each vocabulary entry is always a pre-terminal symbol.
  • the pre-terminal symbol means a symbol which specifies a part of speech immediately before a terminal symbol, while the terminal symbol the last literal term. For example, in the example LD1 of Fig. 8, V is a pre-terminal symbol. Examples of the vocabulary entries are shown in Fig. 8.
  • the features "TENSE”, “NUM” and “PERS” indicate tense, number and person, respectively.
  • the feature values “SG”, “PL” and “3” indicate “singular number”, “plural number” and “third person”, respectively.
  • "V” indicates that "sees” is a verb.
  • the value of the feature "PRED” is put in the symbol ' ' and the portion put in ' ' is called a semantic form.
  • the semantic form gives the information required to interpret the meaning of the corresponding entry.
  • the sentence analysis is performed by two steps: formation of a C-structure and formation of an F-structure.
  • the formation of a C-structure is accomplished by ordinary sentence structure analysis employing phrase structure rules in the grammatical rules (those shown in, for example, Fig. 7) and parts of speech in the dictionary (that shown in, for example, Fig. 8). NO functional schema is used.
  • an F-structure is obtained from the C-structure by using the functional schema.
  • an operation called unification is utilized.
  • the unification is the operation of assembling the F-structure FS1 of V and the F-structure FS2 of NP into a higher-order F-structure FS3 in accordance with the functional schema of the grammatical rule LR2.
  • the unification processing is not only one kind of sum-set operation but the operation of checking consistency.
  • the operation of checking consistency will be explained below.
  • the thus-obtained F-structure FS5 of the sentence is the semantic structure of the sentence which is finally obtained by the LFG analysis.
  • Fig. 1 is a block diagram showing a sentence structure generating device which is suitable for use with both of the first and second embodiments.
  • the sentence structure generating device is arranged to generate a phrase structure from a feature structure.
  • the apparatus shown in Fig. 1 comprises a grammatical rule storing section 11 for storing grammatical rules for use in generating sentence structures.
  • Each grammatical rule stored in the grammatical rule storing section 11 includes a phrase structure portion, a semantic portion, a condition portion and a message portion.
  • Fig. 12 shows one example of the grammatical rules. In Fig. 12, "@sem” indicates the semantic portion, "@con" the condition portion and @mes the message portion.
  • phrase structure portion describes phrase structure rules such as "SDEC ⁇ NP VP" and "VP ⁇ V NP".
  • SDEC, NP, VP and V indicate a declarative sentence, a nominal phrase, a verbal phrase and a verb, respectively, and these elements are called grammatical categories.
  • the semantic portion describes the manner of propagating feature information from a superordinate category (the grammatical category of the phrase structure rule on the left side thereof) to a subordinate category (the grammatical category of the phrase structure on the phrase structure rule on the right side thereof).
  • the condition portion describes the information required to describe the applying conditions for a corresponding grammatical rule.
  • the message portion describes rules for imposing limitations on the application of a grammatical rule using the subordinate category of the grammar as a superordinate category
  • the device shown in Fig. 1 also comprises a grammatical rule search section 12, a grammatical rule interpretation section 13, a generation section 14 and a working memory 15.
  • the grammatical rule interpretation section 13 interprets a grammatical rule which has been searched in the grammatical rule storing section 11 by the grammatical rule search section 12.
  • the generation section 14 generates phrase structures by utilizing grammatical rules and control of the input/output of information of various kinds.
  • a feature structure representing the meaning of a sentence is initially set in the working memory 15 (serving as a working area) and a phrase structure is constructed on the working memory 15 on the basis of the feature structure.
  • the sentence structure generation process according to the first embodiment utilizes a semantic network as an aid, and the procedure is shown in the flowchart of Fig. 9.
  • the sentence structure generation process according to the second embodiment utilizes a recursive procedure employing a semantic network, and the procedure is shown in the flowcharts of Figs. 18A and 18B.
  • Fig. 9 is a flowchart of the sentence structure generation process according to the first embodiment, which is suitable for use in the sentence structure generating device of Fig. 1 according to the first embodiment.
  • Example 1 refers to the process of receiving a feature structure such as that shown in Fig. 10 and outputting a phrase structure such as that shown in Fig. 11 with reference to the grammatical rule shown in Fig. 12, and
  • Example 2 refers to the process of receiving a feature structure such as that shown in Fig. 15 and outputting a phrase structure such as that shown in Fig. 17 with reference to the grammatical rule shown in Fig. 16.
  • Step S201 a feature structure such as that shown in Fig. 10 is set to a grammatical category "bun" ("bun” is a Japanese term which corresponds to a sentence in English) which is located at the uppermost position of a phrase structure tree.
  • the grammatical rule storing section is searched to obtain a grammatical rule which regards "bun” as a superordinate category (a grammatical category on the left side of a phrase structure rule).
  • the desired rule is shown as a rule R1.
  • Step S203 a decision is made as to the grammar applying condition described in the condition portion of the obtained grammatical rule.
  • Step S203 it is determined that this grammar applying condition is not satisfied, and the process proceeds to Step S204, where it is determined whether or not there is another grammatical rule which regards "bun" as a superordinate category. If no such grammatical rule is found, the process proceeds to Step S205, where it is determined that the generation of a sentence structure has failed. If another grammatical rule is found in Step S204, the process proceeds to Step S206, where the grammatical rule is fetched. The process then returns to Step S203, where it is determined whether or not the grammar applying condition described in the condition portion of the grammatical rule is satisfied.
  • Step S203 it is determined whether or not the grammar applying condition described in the condition portion is satisfied. If it is satisfied, it is possible to apply the grammatical rule.
  • Step S207 the phrase structure portion and the semantic portion of the grammatical rule are executed.
  • the phrase structure portion describes a phrase structure rule
  • the semantic portion describes the manner of "propagation" of feature information from a superordinate category to a subordinate category.
  • the feature information is propagated from the superordinate category to the subordinate category, whereby information on the sentence is adequately distributed.
  • the process of propagating the feature information from the superordinate category to the subordinate category will be explained later in detail with reference to Figs. 10 to 12.
  • Step S208 the message portion is executed.
  • This message portion is provided for imposing limitations on the application of a grammatical rule which utilizes as a superordinate category the subordinate category of the grammatical rule fetched in Step S206. More specifically, the message portion is used to control the process of applying a grammatical rule by utilizing the aforesaid condition portion and the message portion.
  • Step S209 it is determined whether or not there is a grammatical rule which utilizes as a superordinate category the subordinate category of that grammatical rule. If such a grammatical rule is found, the process returns to Step S203, where a decision is made as to the condition portion to see if the grammatical rule is applicable. Subsequently, the above-described process is repeated.
  • Step S209 If it is determined in Step S209 that there is no grammatical rule which utilizes the subordinate category as a superordinate category, the process proceeds to Step S210, where it is determined whether or not the subordinate category represents a pre-terminal symbol.
  • the pre-terminal symbol means a grammatical category, such as N or V of Fig. 11, which has no lower-order grammatical category.
  • a symbol whose order is lower than that of the pre-terminal symbol is called a terminal symbol, which corresponds to "the" or "boy” in the example shown in Fig. 11.
  • Step S210 If it is determined in Step S210 that the subordinate category represents no pre-terminal symbol, the process proceeds to Step S211, where it is determined that the sentence structure analysis has failed. If it is determined in Step S210 that the subordinate category is the pre-terminal symbol, the process proceeds to Step S212, where the desired word is extracted from the feature information which has been propagated to the pre-terminal symbol. For example, if the pre-terminal symbol is DET with the terminal symbol in the feature information being "the”, "the” is extracted and placed under the aforesaid pre-terminal symbol. In Step S213, it is determined whether or not words are determined with respect to all the pre-terminal symbols. If there is any pre-terminal symbol for which a word is not determined, the process returns to Step S212, where a word is determined with respect to the pre-terminal symbol.
  • Fig. 12 is a flowchart showing one example of the grammatical rules which are described in the grammatical rule storing section 11.
  • “@sem” indicates a semantic portion
  • "@con” a condition portion
  • “@mes” a message portion
  • "-" indicates the deletion of a feature structure.
  • "++” indicates the addition of a message.
  • "bun” indicates a sentence, "SDEC” a declarative sentence, "END” a punctuation, "NP” a nominal phrase, “BEP” a BE-verbal phrase, "PRED” a predicate phrase, "VP” a verbal phrase, "DDET” a non-indefinite determiner phrase such as a definite article, "NOMHD” a nominal head, "SREL/NP” a relative clause, “RELPRO” a relative pronoun and "V” a verb. Details of the grammatical rules will be explained later during the process of explaining the processing executed by the present sentence structure generating device.
  • Fig. 13 is a schematic diagram showing a semantic network which corresponds to "The boy who I saw is John.”.
  • Fig. 10 is a diagram showing a feature structure which represents "The boy who I saw is John.”
  • the feature structure is a set including pairs of a feature name and a feature value.
  • [1] in, for example, head [1] indicates the link of the feature structure.
  • the feature structure of Fig. 10 is obtained by conversion of the semantic network of Fig. 13. This conversion is performed with reference to an English dictionary.
  • Step S201 the feature structure of Fig. 10 is set as "bun” (Step S201) to start generation of a sentence structure.
  • the grammatical rule storing section 11 is searched to obtain a grammatical rule which utilizes "bun” as a superordinate category (Step S202), and the grammatical rule R1 whose structure is described as "bun” ⁇ SDEC END is found. Since R1 does not include a condition portion, a phrase structure portion, a semantic portion and a message portion are executed in Steps S207 and S208.
  • the grammatical code of the main verb of "SDEC" is Wv1 and matches the condition portion of R2, whereby R2 is applied.
  • the feature name "syn” is an abbreviation of a syntax code.
  • Wv1 is a grammatical code used in "Longman Dictionary of Contemporary English" and indicates a BE-verb.
  • the feature structure of the feature name "SUB” in the feature structure of the feature name “subcat” in the feature structure of the superordinate is propagated to NP.
  • “Subcat” is an abbreviation of subcategorization, and serves to bundle elements having a grammatically strong connection with a main verb (a BE-verb in this case).
  • the message portion of R2 includes the message description "(NP rel) ⁇ NP ++ SREL. This description indicates that, if NP generated in R2 has a relative clause (rel), the massage "SREL" is added to NP. Since the example of Fig. 10 includes a relative clause, SREL is added to NP.
  • the omission information "/NP" is incorporated in the feature structure as a slash feature and is automatically sent according to the propagation of the feature. When this information is set to NP, ⁇ (null) is output.
  • Example 1 has been explained in conjunction with Figs. 10-13 with reference to the linguistic phenomenon of a relative clause.
  • the present invention can be applied to a wide variety of linguistic phenomena.
  • the following is an explanation, referring to Example 2, of the processing of free cases which are related to a decision as to an adequate order of words which plays an important role in generation of a sentence structure.
  • Example 2 The feature structure of Example 2 is, as shown in Fig. 15, obtained from a semantic network corresponding to "I stop at Kobe to rest.” of Fig. 14. Grammatical rules which are associated with the processing of the feature structure of Fig. 15 is shown in Fig. 16.
  • the processing of the free cases will be explained.
  • the free cases include a purpose case (PUR), a space case (SPA), a cause case, a condition case and so on.
  • PUR purpose case
  • SPA space case
  • Stop which will be explained below involve free case elements such as the purpose case PUR and the space case SPA.
  • the processing of the objective case PUR is initially executed in this example.
  • Fig. 17 shows a phrase structure which is finally obtained from Example 2.
  • This program performs recursive processing and generates a structure whose order is lower than that of a particular phrase structure tree by recursively calling "gen" in "gen".
  • a sub-tree is produced by assembling the return values (sub-trees) and a superordinate category, and the sub-tree is set as the return value of "gen" utilizing the superordinate category as an argument.
  • Step S301 the generation program "gen” is called by utilizing as an argument “bun” which is the top category (highest-order grammatical category) of a phrase structure tree and the feature structure of a sentence.
  • the "gen” denoted by numeral S303 generates a return value by repeating a recursive call in the interior.
  • the return value of "gen” S303 is obtained, it is set as a value named "TREE" in Step S302 (Fig. 18A).
  • the desired phrase structure tree is thus obtained.
  • the "gen” S303 is a recursive processing program for generating a phrase structure tree.
  • Step S304 the "gen” S303 sets a grammatical category passed as an argument as a value named "cat” and a feature structure as a value named FS.
  • Step S305 a grammatical rule utilizing "cat" as a superordinate category is searched for.
  • the grammatical rule search section 12 shown in Fig. 1 is used for that purpose.
  • Step S306 it is determined whether or not there is a grammatical rule utilizing "cat" as a superordinate category. If the superordinate category is found, the grammatical rule is applied in Step S307.
  • Step S307 includes Steps S308-S313 and performs a processing similar to that explained in connection with Steps S203-S209. Specifically, before the grammatical rule interpretation section 13 of Fig. 1 applies the grammatical rule, it is determined whether or not the condition portion of the grammatical rule is satisfied. If it is not satisfied, the process proceeds to Step S309, where it is determined whether or not there is another grammatical rule. If there is no other grammatical rule, the process proceeds to Step S310, where it is determined that the generation of a sentence structure has failed. If there is another grammatical rule, it is applied in Step S311.
  • Step S308 If it is determined in Step S308 that the condition portion is satisfied, the process proceeds to Step S312, where the phrase structure portion and the semantic portion are executed to propagate a feature. Then, in Step S312, a message portion is executed.
  • Step S314 When the application of the grammatical rule is completed in this manner, the process proceeds to Step S314, where "gen" is called by utilizing as an argument each subordinate category of the grammatical rule and a corresponding feature structure.
  • the generation section 14 is characterized by the recursive processing of calling "gen” within “gen” in the above-described manner. In this processing, since there are in general a plurality of subordinate categories, "gen” is repeated by a plurality of times.
  • Step S315 it is determined whether or not return values have been obtained from all "gens” that have been recursively called. If the return values of "gens" relative to all the subordinate categories have been obtained, the return values and the superordinate categories are assembled as one list in Step S316 and are returned as the return value of "gen" which utilizes the superordinate category as an argument. This return value is a sub-tree which has a lower order than the superordinate category.
  • Step S317 it is determined whether "cat” is a pre-terminal symbol such as V (verb) or N (noun). If it is determined that "cat” is not a pre-terminal symbol, the control proceeds to Step S318, where it is determined that the generation of a sentence structure has failed. If it is determined that "cat” is a pre-terminal symbol, the process proceeds to Step S319, where a word is extracted from the feature structure propagated to "cat". In Step S320, a sub-tree constituted by "cat” and the word is returned as the return value of "gen” which utilizes "cat” as an argument.
  • V verb
  • N noun
  • the second embodiment differs from the first embodiment in that, even if a semantic network is not employed, it is possible to generate a phrase structure tree from a feature structure by recursively calling the generation program.
  • the present invention enables processing based on a clear construction utilizing grammatical rules expressed by declarative descriptions. This is because the present invention accomplishes a remarkably simple, systematic generation method of generating a phrase structure tree (phrase structure) by the single principle of "propagation of a feature" from a higher-order node to a lower-order node along the phrase structure tree. Accordingly, it is possible to achieve a phrase structure tree which excels in maintainability and expandability. It is also possible to reliably generate a sentence which matches grammar.
  • the sentence structure generating device is arranged to receive as its input a feature structure represented by a feature serving as a constituent element and then to output a phrase structure.
  • phrase structure rules and corresponding case information are described in the same part, it is necessary to describe them by using two expressions S (NP(A) VP(V(*) NP(R) NP(0))) S (NP (A) VP(V(*) NP (0) NP (C)))
  • a phrase structure rule and corresponding case information are described in a phrase structure part and a phrase structure part, respectively. Accordingly, the phrase structure rule can be represented by a single description and the clearness of the construction can be remarkably improved.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Machine Translation (AREA)

Claims (8)

  1. Vorrichtung zur Erzeugung von natürlichsprachlichen Sätzen zur Erzeugung einer Phrasenstruktur aus einer Satzeingabe in die Vorrichtung, wobei die Vorrichtung folgendes umfaßt:
    Eine Grammatikregelspeichereinrichtung (11) zum Speichern einer Vielzahl von grammatikalischen Regeln,
    eine Sucheinrichtung (12) zum Suchen grammatikalischer Regeln, die in der Grammatikregelspeichereinrichtung gespeichert sind,
    eine Interpretationseinrichtung (13) zum Interpretieren der grammatikalischen Regeln, die durch die Sucheinrichtung gesucht wurden, und
    eine Erzeugungseinrichtung zum Erzeugen einer Phrasenstruktur für den Satz,
    dadurch gekennzeichnet, daß die Vorrichtung so gestaltet ist, daß sie einen Eingabesatz verarbeitet, der als Merkmalstruktur bereitgestellt ist, in der die Satzelemente als Merkmale dargestellt sind, und daß jede der grammatikalischen Regeln folgendes einschließt: einen Phrasenstrukturteil, der eine Phrasenstrukturregel beschreibt, die eine grammatikalische Funktion darstellt, einen semantischen Teil, der eine Art und Weise beschreibt, in der Merkmalinformationen in der Phrasenstrukturregel von einer bestimmten übergeordneten Kategorie auf eine untergeordnete Kategorien weitergereicht werden, einen Bedingungsteil, der eine Bedingung beschreibt, unter der die grammatikalische Regel angewendet wird, und einen Nachrichtenteil, der eine Nachricht zum Einbringen von Begrenzungen auf die Verwendung der Phrasenstrukturregel beschreibt, wenn die untergeordnete Kategorie eine neue übergeordnete Kategorie wird, wobei die Sucheinrichtung so gestaltet ist, daß sie die grammatikalischen Regeln gemäß der Art und Weise durchsucht, die im semantischen Teil beschrieben ist, um so die Merkmalinformationen in der Phrasenstruktur weiterzureichen, und die Erzeugungseinrichtung so gestaltet ist, daß sie eine grammatikalische Regel, die durch die Interpretationseinrichtung interpretiert wurde, auf den Eingabesatz anwendet und die Phrasenstruktur aus der Information über die grammatikalische Funktion, die durch die grammatikalische Regel eingebracht wird, erzeugt.
  2. Vorrichtung nach Anspruch 1, worin die Merkmalstruktur des Eingabesatzes beschrieben wird unter Verwendung einer Merkmalbeschreibungsregel einer Vereinheitlichungsgrammatik (unification grammar).
  3. Vorrichtung nach Anspruch 2, worin die Einrichtung zur Speicherung der grammatikalischen Regel so gestaltet ist, daß sie eine grammatikalische Regel speichert, die unter Verwendung der Merkmalbeschreibungsregel beschrieben wird.
  4. Vorrichtung nach einem der Ansprüche 1 bis 3, umfassend eine Einrichtung zur Umwandlung der Merkmalstruktur des Eingabesatzes aus einem semantischen Netzwerk.
  5. Vorrichtung nach einem der Ansprüche 1 bis 3, zusätzlich umfassend eine Analyse einrichtung zum Analysieren eines Satzes, um die Merkmalstruktur des Eingabesatzes bereitzustellen.
  6. Vorrichtung nach einem der vorhergehenden Ansprüche, worin der Phrasenstrukturteil eine Phrase als eine Kategorie einschließt und eine übergeordnete Kategorie und wenigstens eine untergeordnete Kategorie aufweist, die die übergeordnete Kategorie bildet.
  7. Vorrichtung nach einem der vorhergehenden Ansprüche, worin die Merkmalstruktur, die auf die untergeordnete Kategorie weitergereicht wird, wenigstens eine Merkmalstruktur der übergeordneten Kategorie umfaßt.
  8. Vorrichtung nach einem der vorhergehenden Ansprüche, worin die Sucheinrichtung so gestaltet ist, daß sie eine Kategorie, die die Bedingung erfüllt, die im Bedingungsteil der Nachricht beschrieben ist, als untergeordnete Kategorie auswählt, wenn die Nachricht, die im Nachrichtenteil beschrieben ist, von der übergeordneten Kategorie auf die untergeordnete Kategorie weitergereicht wird.
EP90302676A 1989-03-14 1990-03-13 Natursprachenverarbeitungssystem Expired - Lifetime EP0388156B1 (de)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP63253/89 1989-03-14
JP1063253A JPH02240769A (ja) 1989-03-14 1989-03-14 自然言語文生成装置

Publications (3)

Publication Number Publication Date
EP0388156A2 EP0388156A2 (de) 1990-09-19
EP0388156A3 EP0388156A3 (de) 1992-09-02
EP0388156B1 true EP0388156B1 (de) 1997-09-03

Family

ID=13223906

Family Applications (1)

Application Number Title Priority Date Filing Date
EP90302676A Expired - Lifetime EP0388156B1 (de) 1989-03-14 1990-03-13 Natursprachenverarbeitungssystem

Country Status (4)

Country Link
US (1) US5101349A (de)
EP (1) EP0388156B1 (de)
JP (1) JPH02240769A (de)
DE (1) DE69031354T2 (de)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8706477B1 (en) 2008-04-25 2014-04-22 Softwin Srl Romania Systems and methods for lexical correspondence linguistic knowledge base creation comprising dependency trees with procedural nodes denoting execute code
US8762130B1 (en) 2009-06-17 2014-06-24 Softwin Srl Romania Systems and methods for natural language processing including morphological analysis, lemmatizing, spell checking and grammar checking
US8762131B1 (en) 2009-06-17 2014-06-24 Softwin Srl Romania Systems and methods for managing a complex lexicon comprising multiword expressions and multiword inflection templates

Families Citing this family (53)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH03129469A (ja) * 1989-10-14 1991-06-03 Canon Inc 自然言語処理装置
US5418716A (en) * 1990-07-26 1995-05-23 Nec Corporation System for recognizing sentence patterns and a system for recognizing sentence patterns and grammatical cases
US5369574A (en) * 1990-08-01 1994-11-29 Canon Kabushiki Kaisha Sentence generating system
EP0494788B1 (de) * 1991-01-11 1996-10-02 Canon Kabushiki Kaisha Fehlerdiagnose durch Prozess-Simulation
US5297040A (en) * 1991-10-23 1994-03-22 Franklin T. Hu Molecular natural language processing system
JPH05151260A (ja) * 1991-11-29 1993-06-18 Hitachi Ltd 翻訳テンプレート学習方法および翻訳テンプレート学習システム
JP2801459B2 (ja) * 1992-02-21 1998-09-21 富士通株式会社 オブジェクトネットワークによる言語処理システム
JPH05324702A (ja) * 1992-05-20 1993-12-07 Fuji Xerox Co Ltd 情報処理装置
US5369573A (en) * 1992-07-02 1994-11-29 Docustructure Corporation Structured document syntax management
US5675815A (en) * 1992-11-09 1997-10-07 Ricoh Company, Ltd. Language conversion system and text creating system using such
US5384702A (en) * 1993-09-19 1995-01-24 Tou Julius T Method for self-correction of grammar in machine translation
US5717913A (en) * 1995-01-03 1998-02-10 University Of Central Florida Method for detecting and extracting text data using database schemas
GB2300495A (en) * 1995-04-13 1996-11-06 Canon Kk Language processing
US5721938A (en) * 1995-06-07 1998-02-24 Stuckey; Barbara K. Method and device for parsing and analyzing natural language sentences and text
US6292767B1 (en) * 1995-07-18 2001-09-18 Nuance Communications Method and system for building and running natural language understanding systems
JPH0981569A (ja) * 1995-09-12 1997-03-28 Hitachi Ltd 多カ国対応サービス提供システム
WO1998054724A2 (en) * 1997-06-02 1998-12-03 Harris Corporation Web-based management of an enhanced services platform and method
US5960384A (en) * 1997-09-03 1999-09-28 Brash; Douglas E. Method and device for parsing natural language sentences and other sequential symbolic expressions
ITUD980032A1 (it) * 1998-03-03 1998-06-03 Agostini Organizzazione Srl D Sistema di traduzione a macchina e rispettivo tradsistema di traduzione a macchina e rispettivo traduttore che comprende tale sistema uttore che comprende tale sistema
US6330530B1 (en) * 1999-10-18 2001-12-11 Sony Corporation Method and system for transforming a source language linguistic structure into a target language linguistic structure based on example linguistic feature structures
US7099855B1 (en) * 2000-01-13 2006-08-29 International Business Machines Corporation System and method for electronic communication management
US8478732B1 (en) 2000-05-02 2013-07-02 International Business Machines Corporation Database aliasing in information access system
US6704728B1 (en) 2000-05-02 2004-03-09 Iphase.Com, Inc. Accessing information from a collection of data
WO2001098942A2 (en) * 2000-06-19 2001-12-27 Lernout & Hauspie Speech Products N.V. Package driven parsing using structure function grammar
US6408277B1 (en) 2000-06-21 2002-06-18 Banter Limited System and method for automatic task prioritization
US8290768B1 (en) 2000-06-21 2012-10-16 International Business Machines Corporation System and method for determining a set of attributes based on content of communications
US9699129B1 (en) 2000-06-21 2017-07-04 International Business Machines Corporation System and method for increasing email productivity
US7010479B2 (en) * 2000-07-26 2006-03-07 Oki Electric Industry Co., Ltd. Apparatus and method for natural language processing
US6675159B1 (en) 2000-07-27 2004-01-06 Science Applic Int Corp Concept-based search and retrieval system
US7027974B1 (en) 2000-10-27 2006-04-11 Science Applications International Corporation Ontology-based parser for natural language processing
US7043420B2 (en) * 2000-12-11 2006-05-09 International Business Machines Corporation Trainable dynamic phrase reordering for natural language generation in conversational systems
US7644057B2 (en) * 2001-01-03 2010-01-05 International Business Machines Corporation System and method for electronic communication management
US6950793B2 (en) * 2001-01-12 2005-09-27 International Business Machines Corporation System and method for deriving natural language representation of formal belief structures
US7127402B2 (en) * 2001-01-12 2006-10-24 International Business Machines Corporation Method and apparatus for converting utterance representations into actions in a conversational system
US7257537B2 (en) * 2001-01-12 2007-08-14 International Business Machines Corporation Method and apparatus for performing dialog management in a computer conversational interface
US7085723B2 (en) * 2001-01-12 2006-08-01 International Business Machines Corporation System and method for determining utterance context in a multi-context speech application
US7249018B2 (en) * 2001-01-12 2007-07-24 International Business Machines Corporation System and method for relating syntax and semantics for a conversational speech application
US6766316B2 (en) 2001-01-18 2004-07-20 Science Applications International Corporation Method and system of ranking and clustering for document indexing and retrieval
US7526424B2 (en) 2002-03-20 2009-04-28 Microsoft Corporation Sentence realization model for a natural language generation system
JP2004139427A (ja) * 2002-10-18 2004-05-13 Fuji Xerox Co Ltd コミュニケーション支援システム及びコミュニケーション支援システム、並びにコンピュータ・プログラム
US20040193557A1 (en) * 2003-03-25 2004-09-30 Olsen Jesse Dale Systems and methods for reducing ambiguity of communications
US7346493B2 (en) * 2003-03-25 2008-03-18 Microsoft Corporation Linguistically informed statistical models of constituent structure for ordering in sentence realization for a natural language generation system
US7389230B1 (en) 2003-04-22 2008-06-17 International Business Machines Corporation System and method for classification of voice signals
US20050187913A1 (en) 2003-05-06 2005-08-25 Yoram Nelken Web-based customer service interface
US8495002B2 (en) * 2003-05-06 2013-07-23 International Business Machines Corporation Software tool for training and testing a knowledge base
US8060357B2 (en) * 2006-01-27 2011-11-15 Xerox Corporation Linguistic user interface
KR100691400B1 (ko) * 2006-03-31 2007-03-12 엔에이치엔(주) 부가 정보를 이용하여 형태소를 분석하는 방법 및 상기방법을 수행하는 형태소 분석기
US8346555B2 (en) * 2006-08-22 2013-01-01 Nuance Communications, Inc. Automatic grammar tuning using statistical language model generation
US9286589B2 (en) 2011-10-14 2016-03-15 Caelo Media, Llc Method and system for customizing a project
US9064009B2 (en) * 2012-03-28 2015-06-23 Hewlett-Packard Development Company, L.P. Attribute cloud
US9569425B2 (en) * 2013-03-01 2017-02-14 The Software Shop, Inc. Systems and methods for improving the efficiency of syntactic and semantic analysis in automated processes for natural language understanding using traveling features
US9606980B2 (en) 2014-12-16 2017-03-28 International Business Machines Corporation Generating natural language text sentences as test cases for NLP annotators with combinatorial test design
US11449533B2 (en) * 2018-02-22 2022-09-20 Entigenlogic Llc Curating knowledge for storage in a knowledge database

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5840684A (ja) * 1981-09-04 1983-03-09 Hitachi Ltd 自然言語間の自動翻訳方式
JPH083815B2 (ja) * 1985-10-25 1996-01-17 株式会社日立製作所 自然言語の共起関係辞書保守方法
JPH0833895B2 (ja) * 1986-11-28 1996-03-29 富士通株式会社 機械翻訳システムにおける文生成処理方式
US4868750A (en) * 1987-10-07 1989-09-19 Houghton Mifflin Company Collocational grammar system
US4914590A (en) * 1988-05-18 1990-04-03 Emhart Industries, Inc. Natural language understanding system

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8706477B1 (en) 2008-04-25 2014-04-22 Softwin Srl Romania Systems and methods for lexical correspondence linguistic knowledge base creation comprising dependency trees with procedural nodes denoting execute code
US8762130B1 (en) 2009-06-17 2014-06-24 Softwin Srl Romania Systems and methods for natural language processing including morphological analysis, lemmatizing, spell checking and grammar checking
US8762131B1 (en) 2009-06-17 2014-06-24 Softwin Srl Romania Systems and methods for managing a complex lexicon comprising multiword expressions and multiword inflection templates

Also Published As

Publication number Publication date
DE69031354T2 (de) 1998-01-08
US5101349A (en) 1992-03-31
EP0388156A3 (de) 1992-09-02
DE69031354D1 (de) 1997-10-09
EP0388156A2 (de) 1990-09-19
JPH02240769A (ja) 1990-09-25

Similar Documents

Publication Publication Date Title
EP0388156B1 (de) Natursprachenverarbeitungssystem
EP0424032B1 (de) Natürliche Sprache verarbeitendes Gerät
US5848385A (en) Machine translation system using well formed substructures
JP3009215B2 (ja) 自然語処理方法および自然語処理システム
US6470306B1 (en) Automated translation of annotated text based on the determination of locations for inserting annotation tokens and linked ending, end-of-sentence or language tokens
US7072826B1 (en) Language conversion rule preparing device, language conversion device and program recording medium
EP0449230B1 (de) Übersetzungsverfahren und -system zur Übertragung zwischen zwei Sprechern mit unterschiedlicher Sprache
EP0201324B1 (de) Sprachengenerationssystem
US20010029443A1 (en) Machine translation system, machine translation method, and storage medium storing program for executing machine translation method
EP0398513B1 (de) Verfahren und Vorrichtung zur Übersetzung eines Satzes mit einem durch Trennung gebildeten, zusammengesetzten Wort
Uchida 4• 2 atlas ii: A machine translation system using conceptual structure as an interlingua
Erbach Syntactic processing of unknown words
WO1997048058A1 (en) Automated translation of annotated text
WO1997048058A9 (en) Automated translation of annotated text
Sengupta et al. A morpho-syntactic analysis based lexical subsystem
KR100278095B1 (ko) 원문의복잡도에따라융통성을갖는기계번역시스템
JP3113257B2 (ja) 機械翻訳装置
Agustini et al. Selection restrictions acquisition for parsing improvement
Uchida Atlas
Fordham et al. Parsing with principles and probabilities
JP4033088B2 (ja) 自然言語処理システム及び自然言語処理方法、並びにコンピュータ・プログラム
Alansary et al. IAN: An Automatic tool for Natural Language Analysis
Adorni et al. Production of sentences: a general algorithm and a case study
JP2687334B2 (ja) 機械翻訳装置
Aref et al. English to Arabic machine translation: a critical review and suggestions for development

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AK Designated contracting states

Kind code of ref document: A2

Designated state(s): DE FR GB

17P Request for examination filed

Effective date: 19901231

PUAL Search report despatched

Free format text: ORIGINAL CODE: 0009013

AK Designated contracting states

Kind code of ref document: A3

Designated state(s): DE FR GB

17Q First examination report despatched

Effective date: 19941027

GRAG Despatch of communication of intention to grant

Free format text: ORIGINAL CODE: EPIDOS AGRA

GRAH Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOS IGRA

GRAH Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOS IGRA

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): DE FR GB

REF Corresponds to:

Ref document number: 69031354

Country of ref document: DE

Date of ref document: 19971009

ET Fr: translation filed
PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed
REG Reference to a national code

Ref country code: GB

Ref legal event code: IF02

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: GB

Payment date: 20070319

Year of fee payment: 18

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: DE

Payment date: 20070523

Year of fee payment: 18

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: FR

Payment date: 20070319

Year of fee payment: 18

GBPC Gb: european patent ceased through non-payment of renewal fee

Effective date: 20080313

REG Reference to a national code

Ref country code: FR

Ref legal event code: ST

Effective date: 20081125

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20081001

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: FR

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20080331

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: GB

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20080313